Wireless sensor network is a novel technology emerging from embedded system, sensor technology and wireless networks. The rapid deployment, self-organization and fault tolerance characteristics of wireless sensor networks make them a very promising sensing technique for military, environmental and health applications 1. Wireless communication endowed with numerous advantages over traditional wired network and enables to develop small, low-cost, low power and multi-functional sensing devices. These small sensing devices have the capabilities of sensing, computation, self organizing and communication and are known as “sensors”. Sensor is a tiny device used to sense the ambient condition of its surroundings, gather data, and process it to draw some meaningful information which can be used to recognize the phenomena around its environment. These sensors can be grouped together using mesh networking protocols to form a network communicating wirelessly using radio frequency channel. The collection of these homogenous or heterogeneous sensor nodes is known as wireless sensor network (WSN) 2. The ability of low cost, small size and easy deployment of the sensor nodes make it possible to deploy them in a large area to be investigated 3. Interestingly, unlike other networks that performs poor with growth in their network size, WSN get stronger and performs better as much as number of nodes exceeds. In addition, without any complexity in configuration, network size can be extended simply by adding additional number of nodes. Therefore, it is said that connectivity using mesh networking will occupy any possible communication path in search of destination using node to node hoping.
These networks are fast gaining popularity because they have potentially unlimited applications in future. WSNs have attracted a huge amount of attention in many fields including environment, military, health monitoring, disaster alert, battlefield surveillance, habitat monitoring, automobiles, traffic control, building and mining industries. They also have a great potential to revolutionize many aspects of our lives.
WSN is the collection of a large number of spatially distributed autonomous devices cooperatively monitoring an area or phenomena and sending the collected data to a command center using wireless channels. By considering WSNs application domain, one can presume it like a traditional wired or wireless network. But the reality is very different because traditional wired or wireless networks have enough resources like unlimited power, memory, fixed network topologies, enough communication range and computational capabilities [4, 5]. But on the other side, WSNs have a resource constrained nature with respect to energy, computational capabilities and memory resources [6, 7]. WSN has some unique characteristics such as large scale of deployment, mobility of nodes, node failures, communication failures and dynamic network topology. Unfortunately despite these constrained resources we have the same expectation from the WSNs as that from the traditional computer networks.
The resource constrained nature of WSNs impels numerous challenges in its design and operations degrading its performance. These challenges include significantly communication management, unattended operational nature, network lifetime and faulttolerance 8. Therefore, on one side, to improve WSNs performance these challenges are subjected to be investigated. While on other side, the performance of WSN can be achieved significantly by efficient resource utilization. Resource utilization can be enhanced by focusing on factors involved in WSN operations. Communication in WSN has certainly the influences on its resources. The communication pattern of WSNs involves node to node, node to sink and sink to node communication. These communications involve optimal route selection, route maintenance, secure communication and other computations to compete with user expectation and ensure network performance 9. Many applications of the WSN require secure communications. However, WSNs are prone to different types of attacks because of wireless connectivity, absence of the physical protection and the unattended deployment, etc. Therefore, the security in wireless sensor networks is extremely important. The cryptographic methods used for secure communications have a great influence on energy consumption, node and network lifetime. Thus, developing and using the efficient security protocol becomes one of the primary concerns in WSNs. According to 10 “The challenges in the hierarchy of detecting the relevant quantities, monitoring and collecting the data, assessing and evaluating the information, formulating meaningful user displays, and performing decision-making and alarm functions are enormous.” Generally, the wireless sensor network operation involve data acquisition and data reporting therefore it has a data acquisition network and data distribution network and a management center responsible for its monitoring and control as shown in Figure 1.1 below.
The fundamental for any WSNs application is based on the integration of modern technologies like sensor, CPU and Radio performing sensing, processing and communication. Therefore it requires better understanding of modern network technologies as well as of WSNs hardware units in order to have an effective WSN. Despites all these challenges, the importance of WSN cannot be neglected due to its diverse application domain 10.
Abbildung in dieser Leseprobe nicht enthalten
Fig 1.1: Wireless Sensor Network 10
The main components of a general WSN are the sensor nodes, the sink (base station) and the events being monitored.
In WSN, every sensor node has capabilities of sensing, processing and communicating data to the required destination. The basic entities in sensor nodes are sensing unit, power unit, processing unit, communication unit and memory unit to perform these operations shown in Figure 1.2 below.
Sensors play an important role in sensor networks by creating a connection between physical world and computation world. Sensor is a hardware device used to measure the change in physical condition of an area of interest and produce response to that change. Sensors sense the environment, collect data and convert it to fundamental data (current or voltage etc) before sending it for further processing. It converts the analogue data (sensed data from an environment) to digital data and then sends it to the microcontroller for further processing. There are different categories of sensors which are available and can be used depending on the nature of the intended operation.
A typical sensor node processor is of 4-8 MHz, having 4KB of RAM, 128KB flash and ideally 916 MHz of radio frequency. Sensors size and their energy consumptions are the key factors to be considered in selection of sensors [11, 12, 13].
From an energy perspective, the most relevant kinds of memory are the on-chip memory of a microcontroller and Flash memory—off-chip RAM is rarely, if ever, used. Flash memories are used due to their cost and storage capacity. Memory requirements are very much application dependent. Two categories of memory are used based upon the purpose of storage are: user memory used for storing application related or personal data, and program memory used for programming the device. Program memory also contains identification data of the device, if present.
For computation and data transmission, the corresponding units in sensor node need power (energy). A node contains a power unit responsible to deliver power to all its units. The basic power consumption at node is due to computation and transmission where transmission is the most expensive activity at sensor node in terms of power consumption. Mostly, sensor nodes are battery operated but it can also scavenge energy from the environment through solar cells [11, 12, 13].
Internet
Abbildung in dieser Leseprobe nicht enthalten
Fig 1.2: Components of sensor node 13
Sensor node has a microcontroller which consists of a processing unit, memory, converters (analogue to digital, ATD) timer and Universal Asynchronous Receive and Transmit (UART) interfaces to do the processing tasks. This unit is responsible for data acquisition, processing incoming and outgoing information, implementing and adjusting routing information considering the performance conditions of the transmission [11,12,13].
Senor nodes use radio frequencies in order to achieve networking. This task is managed by radio units in sensor nodes that use electromagnetic spectrum to convey the information to their destinations. Usually each sensor node transfers the data to other node or sinks directly or via multi hop routing [11, 12, 13].
The sink is an interface between the external (management center) world and computational world (sensor network). It is normally a resourceful node having unconstrained computational capabilities and energy supply. The sink node can be stationary or dynamic. The dynamic sink node has advantages that it reduces the transmission energy consumption of the sensor nodes, which is considered as a major cause for decreasing the node lifetime as it consumes most of the node's energy.
Besides application-specific tasks of a node, the entire network requires an adaptation to dynamical system requirements 14. The development focus changes from the single result of a sensor node to the cumulative result of the network. Consequentially, the following requirements for the design and implementation process of sensor networks arise:
- Sensor networks have to be self-organizing.
- Cooperative processing of tasks should lead to more precise results and new application fields.
- Sensor networks require security mechanisms that are adaptive to environmental conditions.
- All algorithms and protocols must be energy optimized.
Generally, operation of WSN involves communication between sensor node and the sink (base station). The sensor node senses environment, perform some computation (if required) and report gathered information to the sink node. If sink node is connected with some actuator, it triggers the alarm for human intervention in case of an event of interest 15.
Although sensor nodes are identical devices but their characteristics varies with the network structures. Sensor deployment, coverage, transmission power, computation, reporting, addressing and communication pattern greatly affects the security protocol operation both at nodes and at base station (sink). The following are the types of communications in WSN which support unicast (one-to-one), multicast (one-to-many) and reverse-multicast (many-to-one) in the following ways 15.
i) Node-to-Node: In a multihop communication, data needs to be passed through intermediate nodes in order to reach to destination. Node to node communications is used to pass data from one node to other till it reaches the destination.
ii) Node-to-sink (base station): When sensor nodes want to send responses back to the sink, this communication pattern is used. This is a reverse-multi path communication which means that more than one node can communicate to base station directly or indirectly. This communication pattern can also be unicast if there are multiple base stations or there is a special node (group leader), who is responsible to gather sensed information and transmit it to base station 15.
iii) Sink (base station)-to-Node: This type of communication is required when base station wants to request data from nodes. Typically, the mode for communication is multicast (one-to-many) which means any sensor node having the requested data can respond to the base station. This pattern of communication can also be multicast or unicast if the identification of nodes is unique by their IDs or locations etc 15.
When compared to other categories of wireless networks, wireless sensor networks possess two fundamental characteristics: multi-hop transmission and constrained energy sources. First, since sensor nodes have limited transmission ranges and organizes themselves in an ad-hoc fashion, two wireless sensor nodes that cannot reach each other directly rely on other sensor nodes to relay data between them. In general, data packets from the source node need to traverse multiple hops before they reach the destination. Second, since sensors are usually small and inexpensive, they are assumed to have constrained energy sources, and any protocols to be deployed in sensor networks need to be aware of energy usage. These two characteristics have important implications to the fundamental performance limits of wireless sensor networks. With respect to the performance of wireless sensor networks, the data transmission capacity and the lifetime of the sensor networks are critical and influential towards the design of optimal deployment strategies of these sensor networks. In a wireless sensor network, a group of wireless nodes spontaneously form a network without any fixed and centralized infrastructure. When two nodes wish to communicate, intermediate nodes are called upon to forward packets and to form a multi-hop wireless route. Due to possibilities of node mobility, the topology is dynamic and routing protocols are proposed to search for end-to-end paths. The network nodes rely on peers for all or most of the services needed and for basic needs of communications. Due to the lack of centralized control and management, nodes rely on fully distributed and self-organizing protocols to coordinate their activities. In both scenarios, distributed protocols need to accommodate dynamic changes at any given time because:
- a node may join or leave the network arbitrarily
- links may be broken, and
- nodes may be powered down as a result of node failures or intentional user actions.
Sensor can be classified on the basis of different aspects, including technological aspects, detection means, their output signals and sensor materials and field of application. Although different classification is needed when looking on its application side but can be categorized into following classes 16.
Active sensors stimulate the environment in order to do the measurements. For example seismic sensors, laser scanners, infrared sensors, sonar's and so on 17.
These sensors can monitor the environment without disturbing the environment. Examples of these sensors are: thermometers, humidity sensors, light sensors and pressure sensors etc 17.
Such type of passive sensors require a clear direction in order to measure the environment (medium) e.g. camera and ultrasonic sensors 17.
According to 17, wireless sensor networks can be deployed for various types of applications based on its data delivery requirement, application type and application objectives. The demands of applications vary according to application nature. Some applications are more interested in only data collection but not in robust delivery while in some other applications, delay cannot be tolerated. There are different application classes with different transmission demands. These application classes with different delivery requirements make both software and hardware design of WSN more challenging. Therefore, it is required to classify WSNs applications in classes in order to understand their nature and requirements. Generally, WSN applications can be classified into following four classes:
This class of WSN application consists of sensor nodes which are used rarely. These sensor nodes are inactive most of the time and come to life (active) when a certain event occurs. When the event is detected, individual node sends event report to the sink which may contain some information about the nature of the event and location. The application nature is sensitive in terms of reliability and delay. As soon as an event is detected, WSN reports to sink within no time. A major challenge in this kind of network at application level is to minimize false reporting of the event. Also routing of event to the sink is a design issue from networking point of view. Examples of such applications are [18, 19, 20 and 21].
- Intruder detection in military surveillance
- Quality check at product line/ anomalous behavior
- Detection of forest fire/ Floods
- Seismic activity detection
- Detection of ocean environment
The functional behavior of sensor nodes in these applications is of continuous nature. In these applications continuous monitoring of some activity is recorded and sent to the sink individually like point-to- point communication. But in case of large network, sink is more interested in distributed computation on gathered data rather than individual node reading in order to avoid traffic volume at sink. Sometimes these sensors can be attached with actuators. The sink might need to store the geographical information of the sensor nodes in the area of interest. Monitoring of humidity in a glass house is an example of such applications. Crucial requirement of these applications is efficient utilization of energy. Examples are; [18, 19, 20 and 22]
- Monitoring humidity, temperature and light etc
- Environmental conditions monitoring
- Home/office smart environments
- Health applications
Abbildung in dieser Leseprobe nicht enthalten
The applications in this class also have the additional feature of sink querying besides monitoring. In this case sink has the ability to send a query to a group of sensor nodes for their reading rather than the periodic reporting of the individual node. This allows the sink to gather information of different locations and also helps in validity of the measurements in order to take a decision (trigger an actuator or raise an alarm). Examples of these applications are; [18, 20, 23].
- Environmental control in buildings
- Soil condition monitoring
- Biological attack detection
- Weather monitoring
- Fire alarming
This class of WSN applications consist some of the characteristics of the previous three classes. Tracking applications involve both the detection as well as location information. When a target is detected at any location by a sensor node, it has to notify the sink promptly where accuracy is the main concern. Now, the sink may require initiating queries to the specific set of sensor nodes in order to get the location information of the target. It also helps to verify the measurements of that individual node about the target detection. The decision of triggering actuator or raising an alarm for human intervention is based on the readings received by this set of sensor nodes. Examples of these applications are; [11, 12, 24].
- Targeting in intelligent ammunition
- Tracking of doctors and patients in hospital
- Tracking of inhabitant in a building
- Tracking of animal in forest
- Tracking and controlling the people in park and building
Wireless sensor networks find their applications in a variety of fields where data confidentiality is a major concern. Amongst the different authentication and confidentiality schemes for information security in wireless sensor networks, symmetric cryptosystems are increasingly becoming popular. Most of these schemes focused on providing strong security and paid little attention on their energy consumptions. Our motivation behind this project is to implement a simple, highly secure novel algorithm for the resource constrained wireless sensor networks that will provide better security and prolongs the lifetime of WSN through efficient energy utilization as most of the devices have limited battery life. Here to avoid much complex energy utilizing computations as in various proposed protocols, we make use of simple computations to get prolonged lifetime of sensor nodes and by making the key as dynamic, we get better security. Then, energy utilization at each node and also node lifetime are calculated which are then compared with the results of existing protocols to verify whether the results are improved or not.
The purpose of such schemes is to ensure that the critical information is accessed only by the legitimate user, and not anyone else. Our scheme has several advantages over the existing protocols and improves the wireless sensor network security system. It avoids the use of high storage requirements for keys and the communication messages for rekeying. In addition, one unique advantage is that it increases the lifetime of the sensor nodes by a considerable amount of time.
The chapter wise organization of this thesis is as follows. Chapter 2 deals with the Literature Survey, which comprises of the approaches and work that has been carried out so far in the field of wireless sensor network security and also describes the problem statement in detail by emphasizing on protocol's suitability. Security issues, considerations and challenges identified from literature are presented in chapter 3. Proposed scheme, application scenarios and implementation details are presented in Chapter 4. In chapter 5, the simulation model, results and the related discussion is presented. The results and also compared with the results of existing protocols. Finally, chapter 6 concludes the thesis by presenting conclusion and directions for future research.
In the recent years, wireless sensor network security has been able to attract the attentions of a number of researchers around the world. In this section, various security schemes proposed or implemented so far in wireless sensor networks are reviewed.
SIA 25 from Sensor Network defines a method of how to query from sensor and collect the information from it. Sometimes remote users give query to return the appropriate result(s) of all raw collected data of each sensor, so sensor does not return the actual needed query results and returns all the raw affected data. Therefore, it is necessary to process all the queries locally and send back to remote users. It requires some specific setting for dedicated nodes in the sensor network, called aggregators 25. According to author, he proposed a framework which is called "aggregate-commit-prove 25 ”. In this framework, the aggregate does not only perform the correct task but also provides the confidentiality of “prove of valid task”. To encrypt the sensed information, it uses commit approach.
Some weaknesses of this proposed framework is Denial-Of-Service (DOS) attack and stealthy attack. In stealthy attack, an adversary creates false perception of results to show the user as valid information although this result is not correct according to the measured values. The strength of this method is to calculate and find out the relevant information queried by users. The proposed framework is efficient for information gathering and verification.
This method requires large memory capacity and consumes significant energy for its operation and hence shortens lifetime of sensor node by a considerable period of time. The weakness of this method is also the availability of low resources for sensor such as memory, computation capability and low power capacity. Scalability is another issue for wireless based networks when network grows gradually. In 26, another scheme was provided that was better than SIA.
In 26, author proposed a model for scalability. This model is especially designed for security and reliability of sensor network. The primary focus is on the scalable security of sensor network topology and also presents some core sub networks with protection whenever tampering occurs in it. The scalable security model is divided into two categories (a) Core Sensors and (b) Additional Sensors. These two security models are used for scalability purposes.
Threats can be of two types: human and environmental. The vulnerability of this proposed method according to the author is that the threats can generate in gathered information of an organization. This kind of threat attack is to take information during processes. The strength for these kinds of attacks is: one should properly define the security tier of threats and should calculate the probability level of accidence. Some major weaknesses are: events, physical interface, key management, correct information gathering, hardware and software issues etc.
The strength of core sensors has high computation, high bandwidth and high battery capability. The major weakness of additional sensors is: limited computation power, low battery, short distance range and small in size as compared to the core sensor. This model is not efficient because it creates a lot of overhead during authentication, key management etc which demands large memory and node energy for sensors.
IEEE 802.15.4 sensors define “both physical (PHY) and Medium Access control (MAC) specification 27”. MAC protocol is placed on the top of physical layer. The principal subject of this adopted protocol is how to access the path out of the available paths throughout the network without any misbehave. The most important goal of MAC is to focus on energy efficiency in sensor networks. Sensor network MAC in the recent study indicates that there is no clear converging style available towards best solution of medium access for sensor networks. One of the early encouraging steps towards sensor network architecture is Polastre et al. Sensornet Protocol (SP) 28. This SP occurs between network layer and data link layer and provides an acceptable and standardized interface for MAC protocol. The basic function of SP is to process the data which is occurred at each node and not at the end points. The strength of MAC protocol during communication is:
- Must filter all incoming messages
- Must be guaranteed with flawless messages
- Must detect congestion in advance when it is Occurring
- Must schedule to avoid overlapping
The vulnerability of MAC at data link layer is to waste energy such as collision and wastage of bandwidth, packet overhead. According to security issues, MAC must provide protection against eavesdropping and malicious threats. This is the strength of MAC. Recent work 29 has shown that the energy consumption using MAC is very high when nodes are in idle mode. This is mainly due to idle listening and hence results in reducing the lifetime of sensor nodes by significant time.
Tiny Sec 30 protocol requires low computation power that is available for sensors. This protocol is freely available in TinyOS. This is platform independent and developers can easily use it for sensor network applications.
In end-to-end system delivery of packets, there are weaknesses due to some security threats such as Denial Of Service (DOS) attack. Integrity of a message is checked at the destination point to know the status of the packet whether it has reached safely or not, if the adversary has injected some packets before reaching to the destination so this kind of attack is just wasting the resources of network such as bandwidth and power consumption of a sensor node. The strength of this protocol is to detect the unauthorized and injected packets before reaching to destination at the link layer. TinySec gives two security protocols (i) “Authenticated encryption (TinySec-AE) (ii) authentication only (TinySec-Auth) 30”. In the first case, the actual data is encrypted and authenticates with the MAC. In the second case, the full packet is authenticated with MAC but the actual/payload data is not encrypted.
The limitation of this method is that the existing schemes such as IPSec, SSL/TLS and SSH are inadequate because of high weight during processes.
The Localized Encryption and Authentication Protocol 31 was first proposed by Zhu et al., as a key management protocol for sensor networks designed to support in-network processing. LEAP solves the problem of key distribution and restricts the impact of a compromised node to the network. LEAP uses four types of keys for each node and communication type, they are: “an individual key shared with the base station, a pairwise key shared with another sensor node, a cluster key shared with multiple neighboring nodes, and a group key that is shared by all the nodes in the network”.
In Individual key distribution, each node has assigned a unique key that is shared with pairwise to the base station for secure communication between Base station and node.
In Pairwise key distribution every node shares its pairwise key with its each immediate node in network. This pairwise key is used for privacy and source authentication. Cluster keys are smaller than group keys. This cluster key is shared with each node and each node should share this key with its all neighbors. The purpose of key is to secure the message broadcast locally.
Group keys are globally defined where each group member node can access and share its messages. This kind of key is used between Base station and group member nodes. The Base station simply broadcast the message to the group nodes.
Main drawback of LEAP is the excessive number of messages that must be exchanged during the establishment of keys, which result in increased communication cost as well as energy consumption.
NEKAP 32 is a link layer key management protocol that establishes two kinds of keys: pairwise keys, for link layer pairwise communication; and cluster keys, for link layer broadcast communication. It is similar to LEAP; however, NEKAP is more resilient to node tampering and is even more energy-efficient. In NEKAP, each node pre-loaded with a master key, broadcasts the master key to its neighbors encrypted with a global shared key. The node keys are generated from the master keys of neighbor nodes, making the discovery of these keys more difficult for adversaries. To establish all of the keys, each node broadcasts only three messages, so the protocol is very energy-efficient. Since the key is valid only within its neighborhood and since the impact of a compromised node key can be restricted to the node's neighborhood, these results render the NEKAP protocol intruder resilient.
The main contribution of NEKAP is a key agreement in which each key is valid only in its neighborhood and therefore the impact of a compromised node key can be restricted to that node's neighborhood. Thus, it is impossible for an adversary to carry out a wide-scale attack by capturing only a few nodes. Moreover, the energy cost of this solution is lower than that of previous solution. NEKAP has many advantages for WSNs, because it is intruder resilient and energy efficient.
Unfortunately, NEKAP is vulnerable to replay attacks 33 because of the key establishment process, which includes only three broadcast messages. A malicious node may transmit an old message that was originally broadcasted from a legal node to its neighbor nodes, and the message cannot be authenticated because these two nodes can't communicate directly. Therefore, a malicious node may gain legal status by cheating the chosen legal nodes by transmitting the old message, and then an adversary may launch other attacks, such DOS attacks, black-holes attacks, or masquerade attacks.
In addition, both LEAP and NEKAP are suffering from node tampering during network initialization. The problem stems from an irrational assumption of LEAP. Although NEKAP seems better than LEAP, the problem still exists. Since NEKAP involves the establishment of keys among nodes, thus a number of communicative messages are to be exchanged which reduces node lifetime by a considerable time as 80% of node energy is used for communication tasks.
MiniSec 34 is another security protocol. The strength of this protocol is to secure the network layer. This is the first and generally implemented security protocol with Telos sensor motes. It provides both the facilities like low energy consumption and high security. It has two operating methods, single source based communication is the first method while the second method is based on broadcast multi-source communication channels. This single-source communication is called Unicast (MiniSec U) communication while multi-source broadcast communication is called Broadcast (MiniSec B). Both security schemes use OCB (Offset Code Book) method to give authenticated encryption. The working order of OCB scheme is as under:
[...]
